An analysis of the MIMO-SDMA channel with space–time orthogonal and quasi-orthogonal user transmissions and efficient successive cancellation decoders

We consider space–time transceiver architectures for space-division multiple-access (SDMA) fading channels with simultaneous transmissions from multiple users. Each user has up to four transmit antennas and employs a space–time orthogonal or a quasi-orthogonal design as an inner code. At the multiple-antenna receiver, efficient successive group interference cancellation strategies based on zero-forcing or minimum mean-square error (MMSE) filtering are employed in some fixed or channel-dependent order. These strategies are efficient in the sense that they exploit the special structure of the inner codes to yield much higher diversity orders than would be otherwise possible, while at the same time preserving what we call the decoupling property of the constituent inner codes which enables the use of low-complexity outer encoders/decoders for each user. Motivated by the special structure of the effective channel matrix induced by the inner codes, we obtain several new distribution results on the QR and eigenvalue decompositions of certain structured random matrices. These results are the key to a comprehensive performance analysis of the proposed multiuser transceiver architectures including the characterization of diversity–multiplexing tradeoff (DMT) curves and exact per-user bit-error rates (BERs) without making simplifying assumptions about error propagation.